Group reconciliation fails during Controller Restart with ovs2.3.2
Description
Environment
Operating System: Other
Platform: Other
Attachments
1
- 02 Apr 2016, 03:05 PM
duplicates
Activity
Show:
Abhijit K November 27, 2017 at 5:04 PM
Shuva - any updates on this?
Abhijit K October 30, 2017 at 4:19 PM
We will need to have Shuva update this bug.
Shuva Kar April 5, 2016 at 3:43 AM
No we dont react to the error message as of now
Anil Vishnoi April 4, 2016 at 10:36 PM
I think that's expected behavior and we should ignore the error message if it's OFPGMFC_GROUP_EXISTS. BTW, do we react to the error message?
Shuva Kar April 2, 2016 at 3:05 PM
please apply the following filter openflow_v4.type == 15 or look at packets 182,184,186,188,192,194,
Done
Details
Details
Assignee
Anil Vishnoi
Anil Vishnoi(Deactivated)Reporter
Shuva Kar
Shuva KarExternal issue ID
5646
Components
Fix versions
Priority
MediumCreated April 2, 2016 at 2:56 PM
Updated September 27, 2021 at 9:01 AM
Resolved May 2, 2019 at 12:46 AM
Group Reconciliation fails on a controller node restart with ovs2.3.2 switch
Steps to reproduce:
1. Configure a group(either FF/Indirect/ALL/SELECT) on the controller using POSTMAN
2. Connect an ovs2.3.2 switch. The switch reflects the flow due to reconciliation
3. Start the wireshark on the 6633 port and Restart the controller
4. We see OFPGMFC_GROUP_EXISTS error for the group_mod on the wire.
On analysing the issue, it happens due to the fact that by the Openflow specifications(both 1.3.1 and 1.3.5):
"For add requests (OFPGC_ADD), if a group entry with the specified group identifier already resides in
the group table, then the switch must refuse to add the group entry and must send an ofp_error_msg
with OFPET_GROUP_MOD_FAILED type and OFPGMFC_GROUP_EXISTS code."
(pg 37/47)